Category Archives: Uncategorized

“The Nature of Technology”

There’s been a lot of recent discussion about whether new technology is contributing to our economic problems. It’s a fascinating question, but one that academics are not well prepared to address. Mainstream neoclassical economics generally ignores the question of structural change in the economy and treats technology as a random “black box” variable that is outside of the model (see this book, for example). Historians of technology have written many insightful case studies of certain technologies or certain industries, but as a rule they are loathe to make generalizations. This leaves us without a well-established framework for discussing how technology develops over time and interacts with the economy. Recently, a number of authors have stepped up to address the question using the old idea of “long wave” cycles in economic history (1, 2), and the new field of complexity economics (3).

One particularly interesting book is “The Nature of Technology” by Brian Arthur. Arthur draws from a broad reading of case studies from the History of Technology to argue that the common view of technology is too monolithic, and misses its structure. The book lays out several terms and definitions to clarify the issue. The word technology is split into three terms: individual technologies, domains of technology, and the “Technium”, and Arthur describes the development processes for each category. Below, I lay out a heavily paraphrased and re-organized summary of the argument. Refer to the book for a wealth of concrete examples from history as well as an insightful analysis of the process of invention, that I do not cover in this post. 

First, let’s distinguish three different terms whose meaning is usually rolled into the single word “technology”.


Individual Technologies

The term “individual technologies” refers to devices such as the diesel engine, the smartphone, and the light bulb. It includes complex assemblies such as the passenger jet as well as very simple objects like the ball bearing or the capacitor. It includes hardware as well as software, such as a word processing application. The main distinction to be made is between these individual technologies, and “domains of technology” like electronics or genetic engineering. Domains of technology are collections of related individual technologies and the knowledge around them — whole industries or fields of study —  whereas individual technologies supply a specific functionality: a computer processes information, a circular saw cuts, a bridge carries traffic.

An individual technology is defined as an arrangement that uses one or several natural phenomena as a means to provide a functionality. The natural phenomena can be simple and based on common sense: “wheels roll easily”, “pieces of wood in certain arrangements can support the weight of a seated person”. Or, the phenomena can be more complicated and based on scientific knowledge: “permanent magnets that are mechanically rotated next to a coil of wire will induce an electric current”, the principle behind electric generators.

There are two generalizations that expand this category further. The first is to include not only devices such as the sewing machine, but also processes such as oil refining, or the Haber-Bosch process that is used in chemical factories to produce ammonia. The second generalization is to include “social technologies” such as the joint-stock company or a legal contract. These social technologies use behavioral or organizational phenomena rather than physical phenomena to supply a functionality.

Domains of technology

A domain of technology is a group of individual technologies that are based around a set of related phenomena. For example, the domain of electronics includes components such as resistors, capacitors and transistors that are all used to transmit and modify electric currents. But the domain is more than just a group of individual technologies, it also includes the surrounding knowledge and techniques used to design, build, and use the individual technologies. As such, the domain of civil engineering includes not only heavy-duty cables and bolts for constructing bridges, but also design principles for choosing materials and tolerances.

The Technium

“The Technium” is a term coined by the author Kevin Kelly to refer to technology in the broadest sense. It is the total of all individual technologies and domains of technology: devices, processes, legal systems, infrastructure, etc.


Individual Technologies

Individual technologies vary widely in form, but they share a common architecture. The architecture is abstract, but it is useful for understanding how technology develops over time. At the center of any technology is the main operating principle, the basic scheme for how to achieve the desired functionality. This operating principle forms the backbone of the technology, and it is made up of an arrangement of sub-technologies. The main operating principle is then augmented by a collection of supporting technologies that improve the main technology’s performance or make it more convenient to use. The architecture is depicted in Figure 1.

Figure 1: Individual technologies are made up of a main operating principle built from several sub-technologies, as well as a set of supporting technologies that enhance and expand its performance. Each sub-technology and supporting technology is a technology in its own right so the structure is recursive.

For a concrete example of this architecture, consider the automobile. The main operating principle of a car is to use an internal combustion engine to transform chemical energy into mechanical motion of the wheels, which then move a load across a surface. The sub-technologies that form the main operating principle are the engine that converts the chemical energy into mechanical energy, the wheels that transform the mechanical energy into linear motion, and the car chassis that holds everything together and supports the load. Notice the recursive structure here: the main technology (a car) is made up of sub-technologies (wheels, engine, chassis), and each sub technology is in turn made up of sub-sub-technologies (for the wheel: axles, bolts, rubber tubes).

Figure 2: The main operating principle of the car is built out of the sub-technologies of the engine, wheels, and chassis.

The car is then augmented by numerous supporting technologies. The most important supporting technologies set up the proper conditions needed for the operating principle to function: the fuel tank that supplies the car engine, the cooling system that keeps it from overheating. Additional supporting technologies are added over time to enhance performance and ease of use.  These supporting technologies make up the difference between the earliest automobile prototypes and modern cars. The early prototypes are not much more than the engine, wheels, and chassis; but the modern car is bristling with supporting technology. Computer chips and sensors monitor the car engine and improve its performance, a cabin and climate control system provide comfort and safety for the passengers, headlights allow the driver to operate the car at night.

Figure 3: Henry Ford’s first car, the Ford Quadricycle (top) and the 2013 Ford focus. The two cars share the same operating principle, but the modern car has many more supporting technologies. [images via Wikimedia commons (top) and Ford Motor Company (bottom)]

Domains of Technology

Recall that a domain of technology is akin to an industry or field of study like genetic engineering or electronics. A domain of technology is a toolbox of components and practices. It includes a group of individual technologies as well as the surrounding knowledge of how to build and use them. This knowledge includes both formal knowledge, the kind that is written up in textbooks, as well as the “cookery” of the field. This cookery consists of rules of thumb and techniques that practitioners in the field learn through experience, but are not easily written down as formal knowledge.

An individual technology is designed within a certain domain, using its components and rules of combination (although many technologies are combinations of sub-technologies from different domains). We can get a feel for how this works by considering a functionality that switched domains over the years. The F-1 rocket engine was a huge machine that powered the Saturn V moon rocket. In order to start firing the rocket, a complicated sequence had to be followed that involved opening certain valves to let propellant flow to different parts of the engine, but only once pressures in other parts of the engine had risen to sufficiently high levels. In modern rockets this type of if-then logic for deciding when to open valves is built within the domain of computer chips and digital sensors. However, the Saturn V was designed in the 1960’s when electronic computers were still bulky and unreliable, so the control system was built in a different domain, it used “fluid mechanic logic”: a series of tubes and valves arranged to perform logical operations based on the fluid pressure (4,5).


Different but interrelated processes govern the development of individual technologies, domains of technology, and the interaction between technology and the economy.

Individual Technologies

Individual technologies develop over time in three different ways: internal replacement, structural deepening, and invention.

Internal Replacement

Internal replacement happens when a sub-technology or supporting technology is replaced by a better version of the sub-technology, without otherwise changing the architecture. For example, as computer chips advanced over the years everything from washing machines to airplanes swapped out their electronics for increasingly powerful microchips. Internal replacement also occurs when new materials are developed. For example, the Boeing 787 uses composite materials for many structural components that were built out of aluminum in older aircraft.

Structural Deepening

Structural deepening is the process of accumulating supporting technologies over the years. Initial prototypes of a new technology are built under time and budget constraints and are usually bare-bones, unrefined versions of the basic operating principle. Over the years people and companies refine the original technology by adding supporting technologies. The supporting technologies are added in order to improve performance, get around part limitations, adapt to a wider range of tasks, handle exceptional circumstances, and enhance safety and reliability.


Finally there is invention. Invention occurs when someone develops a new operating principle for a technology. This can happen when an old functionality is provided by a technology designed in a new domain (for example, the electric light bulb replacing oil lamps for room lighting), or when demand for a new functionality arises and is satisfied by an invention.

Domains of Technology

Domains of technology are based around phenomena that are naturally clustered into related families: optical phenomena, chemical phenomena, quantum phenomena. New domains are created when people learn to work with and understand a new cluster of phenomena. This new understanding leads to a train of new technologies that are designed within the new domain. For example, the investigation of electrical phenomena in the early 1800’s lead to the invention of electric lighting (light bulb), electric communication (telegraph and telephone)  and electric transportation (electric locomotive) in the mid to late 1800’s.

Understanding of a new domain is built out gradually, piece by piece with later developments depending on earlier ones.  The domains are built out through a combination of tinkering as well as science. The earliest domains were investigated with tinkering alone; for example, ancient buildings were designed based on rules of thumb gleaned from experience, because a theoretical basis for structural engineering did not yet exist (6). For later domains such as quantum and genetic phenomena, formal scientific knowledge supplemented tinkering.

The development of a domain requires the buildup of new technologies from the same domain (the investigation of electric phenomena depended on the invention and improvement of voltage and current meters) and it also depends on the technologies that have been built up from older domains (the structure of DNA was determined with the aid of X-ray diffraction images — technology from the domain of wave physics).

Technology and the Economy

New technologies come in waves as new domains are built out, and are then adopted into the economy. But the adoption process is not simple or rapid, it plays out over decades as the new domain and established industries in the economy mutually adapt to each other.

Any given industry uses a set of existing technologies to perform its basic operation: spreadsheets for accounting, canals and barges for transporting goods, a system of paper and filing cabinets for storing medical records. When a new domain comes along, it offers a new design space that can be used to solve the old problems of an industry in a new way. Some initial technologies from the new domain can simply be swapped into the old industry, but often it is more complicated and plays out over decades. This is because industries have specialized needs, and new technologies need to be modified to to address these niches. Furthermore, the industries often need to re-organize how they do business in order to take advantage of the new technology. The text provides a concrete example:

Before factories were electrified a century or so ago, they were powered by steam engines. Each factory had a single engine, a giant hissing and cranking contraption with pistons and a flywheel and a system of belts and pulleys that turned shafts on several floors of a building. These in turn powered all the factory’s machinery. Then electric motors— component technologies of the new electrical domain— became available in the 1880s. They were cheaper in energy use and could be installed as multiple small single units, each next to the machine it powered. And they could be controlled separately, switched on and off as needed. They were a superior technology. Why then did it take American factories close to 40 years to adopt them? [Paul] David found that the effective use of the new technology required a different physical construction of the factory than the old steam-engine layout. It required literally that the factory be re-architected. Not only was this expensive, as we saw in the Frankel case, but just how the factory should be constructed was not obvious. The electricians who understood the new domain were not architects, and factory architects were not electricians. So it took considerable time— in this case four decades or more— to accumulate knowledge of how to accommodate factory design to the new technology and for this knowledge to spread.

The mutual adaptation of domains and industries described here is similar to “long wave” descriptions of economic history such as the ones promoted by Carlotta Perez and Michael Lind. In the long wave picture, new domains of technology drive cycles in the economy that play out over 40-60 years. The effects of the new domain spread in waves, gradually being adopted by different industries and ultimately modifying government and institutions. This whole process is mediated by finance, government policy, and the availability of workers skilled in the new domain.


These new terms and definitions set up a framework for understanding how technology evolves and is adopted into the economy. The framework takes what was a vast and inscrutable process and delineates it into a set of more manageable sub-phenomena. The list below summarizes the main processes that drive the development of technology and its integration into the economy.

Are Inventions Inevitable? Simultaneous Invention and The Incremental Nature of Discovery

Thomas Edison invented the light bulb in 1879. What if he had never been born, Would we still have light bulbs? And would they still have been invented in 1879? It turns out that this is not just a philosophical question and the answer is yes, the light bulb would have been invented at roughly the same time. We know this because at least 23 other people built prototype light bulbs before Edison1, including two groups who filed patents and fought legal battles with him over the rights (Sawyer and Mann in the U.S. and Swan in England)2.

Independently invented lightbulbs from Edison, Swan, and Maxim. Image from “What Technology Wants” by Kevin Kelly.

This is not a strange coincidence that happened with electric lighting, it is the norm in both technological invention and scientific and mathematical discovery. Newton and Leibniz independently invented calculus, Alexander Graham Bell and Elisha Gray both filed a patent for the telephone on the same day — within three hours of each other — and sunspots were simultaneously discovered by four scientists living in four different countries. The list of simultaneous independent inventions includes the airplane (2 people), the steamboat (5 people), photography (2 people), the telegraph (5 people), and the telescope (9 people). In science and math it includes decimal fractions (2 people), the theory of natural selection (2 people), the discovery of oxygen (2 people), molecular theory (2 people), and the conservation of energy (4 people)3.

A study by Ogburn and Thomas4 in 1922 produced a list of 148 major inventions and discoveries that were made independently by two or more groups at the same time. A similar study by Merton5 in 1960 led him to conclude that “the pattern of independent multiple discoveries in science is in principle the dominant pattern, rather than a subsidiary one”. Lest you think that only the landmark discoveries covered in these surveys are subject to multiple invention, recent work by Lemley6 suggests that 90-98% of patent lawsuits are filed against independent inventors and not copiers. Even the idea that multiple simultaneous invention is the norm was advanced by multiple independent groups at the same time7.

All this synchronicity reveals something fundamental about the nature of invention. In popular culture inventors are romanticized as lone geniuses working in home laboratories who come up with completely new ideas — things that establishment thinkers initially dismiss as nonsense. But in practice, it just doesn’t happen this way. Most inventors add to a body of work that is built up by many people over decades. They rise to fame when their invention, one among many in a long chain, is a crucial step that finally enables a practical telephone, airplane, or mathematical proof.

In the case of the light bulb, Edison’s main contribution was a new filament made out of a certain species of bamboo that worked better than the carbonized paper used by Sawyer and Mann before him. The incandescent bulb — a resistive filament looped inside an evacuated glass bulb — had already been developed. Filament material continued to be improved upon by others after Edison’s work. Samuel Morse’s patented contribution to the telegraph was the application of efficient electromagnets for boosting the signal in the wire over long distances. The idea of sending messages over wires using electricity already existed in the world. Morse did not even invent these new electromagnets, but built on the work of Joseph Henry8.

Viewed in this light simultaneous invention is not so surprising. As Lemley puts it, “An isolated flash of genius could strike at any time, while the thirteenth step in a multistage inventive process is likely to come after the twelfth.” The incremental nature of invention has another fascinating consequence besides simultaneous invention. Since significant discoveries require a base of knowledge and tools that is outside the scope of any one person to create, new inventions only happen when “the time is right”. Given the knowledge and tools that exist at any one point in history there is only a finite set of inventions that are possible to create. These things lie within the “adjacent possible” — all those new inventions and ideas that are one step removed from what we already have. Steven Johnson describes how this played out for the discovery of Oxygen in 1774, which was done independently by both Priestley and Scheele9:

“To go looking for oxygen, Priestley and Scheele needed the conceptual framework that the air was itself something worth studying and that it was made up of distinct gases; neither of these ideas became widely accepted until the second half of the eighteenth century. But they also needed the advanced scales that enabled them to measure the miniscule changes in weight triggered by oxidation, technology that was itself only a few decades old in 1774.”

The invention of the steam engine was also only possible because of the work built up over prior decades. By the beginning of the 18th century “[t]he nature of the vacuum and the method of obtaining it were known. Steam boilers capable of sustaining any desired pressure had been made. The piston had been utilized and the safety valve invented”. After this, early versions of the steam engine were built by Savery and then Newcomen before James watt improved the engine yet again by adding a separate condenser and became known to history as the inventor of the steam engine10.

Taken together, all this suggests that inventions are inevitable. Once an idea enters the realm of the adjacent possible the supply of people who are smart enough to act on it is sufficiently large that the idea will quickly be discovered, often by multiple people: “given the electric motor and the train, is not the electric train inevitable?”s. Inventions are not pioneering feats of individuals, but the necessary results of a social process. Mix together people, technology, and the laws of physics and then stir for a decade or two. Out will pop a set of inventions that will expand the boundaries of the adjacent possible so that you can repeat the recipe.

This idea is supported by research into the inventions of early humans who were isolated on separate continents. If inventions are inevitable, then we would expect that each civilization would develop roughly the same tools and that they would invent them in the same order. Kevin Kelly summarizes this research in his book “What Technology Wants” 12: “Each technological progression around the world follows a remarkably similar approximate order. Stone flakes yield to control of fire, then to cleavers and ball weapons … The sequence is fairly uniform. Knifepoints always follow fire, Human burials always follow knifepoints, and the arch precedes welding.” When we invent new technology, we are not blazing a trail into the future, we are uncovering a path that was already there.

1. Fridel, Israel, and Finn, “Edison’s Electric Light: Biography of an Invention” via Kelly, “What Technology Wants”.
2. Lemley, “The Myth of the Sole Inventor” via Thompson at The Atlantic.
3. Ogburn and Thomas, “Are Inventions Inevitable?” via Lemley (note 2).
4. Ibid.
5. Merton, “Singletons and Multiples in Scientific Discovery” via Lemley (note 2).
6. Cotropia, and Lemley, “Copying in Patent Law” via Lemley (note 2).
7. Merton (note 5).
8. Lemley (note 2).
9. Johnson, “Where Good Ideas Come From”.
10. Ogburn and Thomas (note 3).
11. Ogburn and Thomas (note 3).
12. Kelly, “What Technology Wants”.

America’s Most Interesting City

As I mentioned in my last post, I am visiting Detroit this week. This city has long been a symbol of urban poverty, racial unrest and industrial decay. But recently there has been another story to tell — one of cheap land and creative opportunity. An article in Pluck magazine describes this new sentiment:

“Who isn’t invigorated by the prospect of buying a house for $10,000? The New York Times gushes about the proliferation of small, locally-owned businesses in Detroit, epitomized by Slows Bar B.Q. Oprah inspires housewives with stories of the city’s urban gardeners. Radicals talk about the prospects for organizing autonomous communities given the availability of cheap land and the lack of regulation.”

I traveled through the city on Monday to see both of its faces, starting with a tour of high tech businesses downtown and ending in the ruins of an early 20th century car factory.

Downtown Revitalization
I started my trip by taking a free walking tour of downtown from the D:hive, a local nonprofit organization that strives to attract and retain young talent in Detroit. They provide information and resources from their downtown storefront to build community and help new transplants find housing and job opportunities. The tour, offered weekly, takes a small group around the central core of downtown to see beautiful historic architecture that has been renovated to house new businesses that are moving into the city. The most high profile recent champion of Detroit is Dan Gilbert, the founder of Quicken Loans. As described in this Forbes article, Gilbert has invested hundreds of millions of dollars in the city, relocating Quicken’s world headquarters to a renovated tower in the center of downtown as well as taking advantage of a “skyscraper sale” (The 37 story David Stott building was on the market for $3.9 million in 2007 ) to purchase and refurbish 2 million square feet of office space to encourage other businesses to follow suit. This space is now nearly fully occupied. Gilbert has also invested in restoring the city’s waterfront and started bizdom to provide seed funding and mentoring to new startups.

The downtown headquarters of Quicken Loans.

In the article, Gilbert explains his motivations for the investments:

”If you want to attract the kind of brains to grow your business — and there are a lot of them around this area, with dozens of great universities — you need a strong urban core. It’s no secret that people in their 20s and 30s want to be in a vibrant, exciting, urban core. We’re not going to get those people if we’re in a nice building in the suburbs.”

This trend of young people seeking city life has been covered in recent reports and magazine articles. An article in The Atlantic describes some surveys by the National Association of Realtors and the Urban Land institute which indicate that around 60% of 20-30 year olds report that they would prefer to live in walkable urban communities. An analysis of census data also highlights the trend, showing that most of America’s largest metropolitan areas grew faster than their surrounding suburbs in 2010-2011, something that has not happened since the 1920’s. These changing preferences are more than just a matter of fashion, they are driven by powerful economic forces in the modern economy. Enrico Moretti describes three such forces in his book “The New Geography of Jobs” : thick labor markets, specialized service providers, and knowledge spillovers.

In an economy fueled by unskilled workers, its not important whether factories are located in small towns or large cities. If a small town has a surplus of workers and not enough jobs then it would be a good place to build a new factory. A suburb close to expanding factories that are facing a shortage of workers would be a smart place for a job-seeker to move. But when skilled workers are involved the size of the job market becomes much more important for both workers and businesses. Moretti provides a particularly apt analogy comparing a skilled job market to a dating website: would you rather join a site with 10 women and 10 men, or one with 10,000 women and 10,000 men? Because daters are looking for very specific attributes in a mate, they have much better chances finding Mr./Ms. Right in a “thick” dating market that has many buyers and sellers. The same is true for modern tech companies seeking skilled workers to fill specific roles.

In addition to favorable labor markets, cities also have an abundance of specialized service providers such as law firms and accountants that know how to deal with the specific problems a specific industry encounters. And to top it all off, studies of patent citations have shown that information spreads more quickly locally through a city than it does to more distant locations, presumably because of the superiority of face-to-face meetings to long distance communication. These economic forces are powerful. Thousands of high tech firms and startups have moved their offices to New York and San Francisco, paying exorbitant rents and high salaries in order to reap the benefits that come from working in a city. But there is a catch for those who hope to take advantage of these trends to revitalize old cities: it’s a chicken-and-egg problem. Cities only provide these benefits if there is a vibrant hub of high tech companies all working in the same place. It is unclear whether Gilbert’s aggressive investments will be able to overcome this barrier and jump-start an innovation hub in Detroit, but there are some promising signs: Twitter opened a new office downtown in April, and hip bars and restaurants are springing up such as the Grand Trunk Pub which is housed in an old railroad station and serves a wide variety of locally brewed craft beer.

Ruins and Opportunity

On the tour I met an artist and her studio manager from New York — Kumi and Erik — who stopped to check out Detroit on their way back from artprize in Grand Rapids. They are the type of people that Detroit revitalizationists hope to lure away from the high rents of NYC. While they were not about to pick up and move, Erik was clearly taken by Detroit’s charms, repeatedly noting “they say that Detroit is the most interesting city in America right now”. In addition to the grand architecture and ghostly beauty of the abandoned buildings, his attraction was driven by a sense of frontier opportunity — the relative scarcity of people and development compared to crowded cities like New York leaves more room to experiment with new ideas.  As the tour guide put it, “In Detroit all you need is a good idea to Succeed. In New York or Chicago, you need a good idea plus 5 million dollars”. Later, driving through the city looking for the ruins of an old high school (it turned out that it was demolished in 2011), Erik hit a rare snarl of traffic that blocked our path down a one-way street.  “fuck it, it’s the wild west out here” he let out as he drove the car in reverse back onto a wide boulevard, devoid of cars.

Our first big stop after the tour was the Heidelberg project, an outdoor art project that started in 1986 when Tyree Guyton began decorating abandoned houses on his block with bright paint and found objects. The project is controversial in the local neighborhood as many neighbors consider it a gaudy eyesore, and the city of detroit partially demolished it on two occasions. But these days it has gained considerable fame outside the city, drawing 275,000 visitors from across the world every year.

One of the decorated abandoned houses that make up the Heidelberg project.

The project is impressive for its sheer size — stretched out across a city block — as well as its clashing aesthetics of childlike joviality and creepy, ghostly decay. In one house, oversized stuffed animals of the type you might win at a state fair are plastered across the outside wall. A monkey and a cuddly teddy bear stare out from behind a broken window, thier colors faded from exposure to the elements.

Stuffed animals peer out the broken window in a house in the Heidelberg project

After walking up and down the block, we struck up a conversation with Tim Burke, a detroit artist who maintains a studio in a house at one corner of the Heidelberg Project. After talking shop with Erik about strategies for an art contest in which they both competed, he invited us into his gallery to show us the spoils of many years foraging the city. The objects were multitudinous and fascinating — flawed gears discarded from a factory making F-15 fighter jets, foam offal from a nearby toy factory that resembles large mushrooms, and records from a dentist’s office abandoned in the 1960’s complete with black and white photo strips of the patients and extracted human teeth. Burke integrates these found objects into his artwork. For him, the city has been fertile ground.

The final stop on our tour was the Packard Automotive Plant, a car factory built in the early 1900s by the famous industrial architect Albert Kahn and notable for its early use of reinforced concrete. The factory closed in 1956 and was turned into a multi-use industrial park until 1999 when it was abandoned. After this, vandals and trash dumpers filled the buildings with graffiti and refuse including several old boats and vehicles, and scrappers went to work dismantling and stripping the buildings. Today, the site looks like a scene out of war-torn Europe in the 1940’s. Sections of the building have collapsed under their own weight and the courtyards overflow with rubble.

A courtyard of the abandoned Packard Automotive Plant.

In most cities a place like this would be fenced off to keep out trespassers but such a thing would be a daunting task in this section of Detroit because it seems like more than half of the buildings in the surrounding area are unoccupied. As we walked through one of the many courtyards, smoke wafted out of a third-floor window. Climbing the stairs to take a look, we found small fires scattered across the concrete floor burning unattended. Fires at the Packard plant are common, and the fire department no longer intervenes.

Small fires smolder on the 3rd floor of a building in the Packard Automotive Plant.


The ruins of the Packard automotive plant.

I am in Detroit this week to explore a city in the midst of transformation. Detroit was the epicenter of the American technology industry in the early 1900s — the Silicon Valley of its day, where mechanical tinkerers came together to start the automobile industry. It has been a bumpy ride since then. In a sense, Detroit was a victim of its own success. Cheap mass-produced cars and an extensive freeway system built in the 1950s and 60s led to a massive outflow of population from the city of Detroit to its suburbs. Today the population is less than 40% of its 1950 peak and the city is littered with vast swaths of abandoned houses and commercial buildings, such as the Packard automotive plant pictured at the top of this post.

Population of Detroit. Data from

But amidst the apocalyptic landscape, there is a growing movement to revitalize the city. Artists are combating the blight of abandoned buildings and storefronts by covering them with colorful installations while high tech companies such as Quicken loans are opening offices downtown and funding organizations to nurture the next generation of tech startups. It is, of course, uncertain whether these revitalization efforts will succeed, but they are buoyed by economic forces that are spurring the growth of cities across the country: in 2011 America’s large cities grew faster than their suburbs for the first time in 100 years.

An abandoned storefront in downtown Detroit is decorated with post-it notes.

Automation and Employment

There has been a lot of media coverage about job automation recently. The interest is sparked by news of rapid technological progress and the unemployment rate which remains stubbornly high nearly 3 years after the end of the recession. Articles in Slate and the New York Times focus on advanced software that is capable of automating tasks performed by highly paid, highly educated workers like screening mammograms for breast cancer or sorting through legal documents. Two books on the subject (Race against the Machine, The Lights in the Tunnel) take a broader view, proposing that technological advances are causing structural unemployment as more and more tasks are automated. One chart from from the authors of Race Against the Machine shows that the ratio of employment to population has decreased by almost 5% since 1995 even as GDP, corporate profits, and corporate investments have all steadily increased (caveat: these changes may be the result of demographic factors and not automation).

There is also some historical evidence that automation can lead to big problems in the economy. An article by Joseph Stiglitz claims that the Great Depression was a result of farm mechanization that eliminated the need for many farm laborers. The problems were ultimately resolved, according to Stiglitz, when massive government spending on factories for the war effort helped transition the surplus farm workers to factory jobs.

Another lesson from history is more optimistic. Technology and social changes repeatedly caused huge shifts in the way we live and work over the past century, but the economy has nimbly adapted each time (this report from the Bureau of Labor Statistics gives a nice overview of occupational changes in the 20th century). In 1850, 80% of the workers in the U.S. worked on farms. Increasingly powerful and sophisticated farm machines, such as this modern combine, drastically reduced the number of farmers needed to feed the country to the point that only 2% of the modern U.S. population works on a farm. There have also been social changes that had a profound impact on the workforce. From 1960 to the mid 2000’s, the fraction of U.S. population that was employed increased from 20% to 30%, largely because women entered the workforce. Despite these and many other changes, the economy provided enough jobs, on average, to keep the country at full employment throughout the century.

So there is reason to be optimistic that the economy will adapt to upcoming technological changes. That is a good thing, because large changes seem to await us in the next few decades. Here is a table of the ten largest occupations by employment as well as technologies that may automate some of their workload in the future.



Number Employed (Millions) Automation Technology
Retail Salespeople


online product reviews and mobile devices (1,2)


self-checkout kiosks (1,2,3)
Office Clerks


continued advancement of office automation
Food Preparation and Serving Workers (Fast Food)


Registered Nurses


Waiters and Waitresses


tablet menus (1)
Customer Service Representatives


speech recognition and automated call centers (1)


robotic floor cleaners (1,2)
Laborers and Material Movers


mobile robots(1,2,3)


Virtual Assistants (1,2)

Source Data

As you can see there are technologies on the horizon that could, in principle, reduce the need for many of the largest occupations. But don’t expect any of these jobs to disappear overnight – there are more than 600,000 bank tellers in the United States even though ATMs were introduced in the 1970’s (Here is a commercial for a bank with an ATM from 1980). Other occupations that have been automated by technology have not disappeared as quickly as some people assume, although they are declining. This includes travel agents (70,000 employed), telephone operators (18,000 employed) and file clerks (212,000 employed).

Big changes in how people are employed are on the horizon because automation will reduce or eliminate the need for many occupations that are common today. These changes will not happen immediately, but will probably play out over a decade or more, and, if history is a guide, the economy will adapt to create new jobs to employ the displaced workers.